Breaking the loop: Why algorithmic dispersion is essential for safer feeds

Ever noticed your social media feed starting to feel like déjà vu? A video you paused on once soon appears again and again. Wherever you may be scrolling, recommendation engines are working behind the scenes to quietly curate a reality that feels tailor-made for you, yet also strangely predictable. This hidden dynamic is known as algorithmic repetition: a subtle feedback loop that can shape your worldview by continually reinforcing certain types of content.

The longer algorithms observe your habits, the narrower your feed becomes, cycling through familiar themes: videos you might like, posts you’re likely to engage with, news you’re expected to read. On the surface, this repetition seems harmless: recipes, fitness tips, travel diaries. But as media psychologist Dr. Jessica Piotrowski explains, even benign content can turn harmful when reinforced too often.

Recently, Dr. Piotrowski, a Professor in the Amsterdam School of Communication Research (ASCoR) at the University of Amsterdam (UvA) where she holds the Chair Communication in the Digital Society, spoke with Ailís Daly, Head of Trust & Safety for EMEA at WebPurify, an IntouchCX company, about the risks of algorithmic repetition and why platforms (and their users) can benefit from considering algorithmic dispersion, or intentionally diversifying content, to create healthier digital experiences.

When repetition becomes perception

Social media platforms engineer recommendation systems to maximize engagement, tracking every interaction — every pause, click, watch and like. And the logic is simple: if you enjoyed something once, you’ll likely enjoy it again. But the psychological impact isn’t so straightforward.

Dr. Piotrowski, who advises stakeholders such as Google, the European Commission, Meta, YouTube, and NASA, refers to this phenomenon as the “sprinkle effect.”

“It’s about the accumulation: the drip, drip, drip effect. Small doses repeated over time can alter how we think and feel.”

To put it into perspective, imagine a teenager who frequently sees influencers promoting extreme diets. Individually, these posts might not violate platform policies, but in aggregate they send a powerful, and possibly harmful, message: this is how you should look, what you should eat, who you should aspire to be.

Gradually, these repeated exposures begin to normalize unrealistic standards, influencing one’s self-esteem and potentially encouraging harmful behaviors.

Algorithms can also misinterpret signals. “Algorithms make assumptions,” Dr. Piotrowski explains. “Just because you watched a video about dieting doesn’t mean you’re endorsing it. You might be researching critically. Yet the algorithm interprets engagement as interest and amplifies similar content, potentially pushing harmful ideas unintentionally.”

In short, the potential problem with algorithmic repetition isn’t necessarily the content; it’s the way, collectively, it can influence and gradually reshape what you see as normal.

The case for algorithmic dispersion

If repetition risks narrowing perspectives and distorting realities, the solution perhaps lies in algorithmic dispersion. Instead of continually amplifying predictable content, dispersion deliberately introduces diverse content into feeds to help break repetitive cycles.

“We need to create experiences that show young people more of the world, not less,” Dr. Piotrowski argues. “Personalization can be powerful, but it must avoid becoming a tunnel.”

She offers an analogy: “Imagine a grocery store. When you walk into a store now, there are aisles with clear labels telling you where to buy fruit, where to buy milk, where to buy bread.

“Without algorithms sorting products into aisles, everything would just be scattered chaotically. Algorithms organize content efficiently, but too much optimization narrows our experiences. The goal isn’t to remove algorithms, but to ensure they provide a balanced, enriching experience.”

Designing feeds that support wellbeing

Platform design choices shape more than just their users’ clicks: they shape beliefs and values. Dr. Piotrowski points out that social media platforms were never intended to be environments where young people spend hours each day, and yet they have become the default spaces for identity formation and social comparison.

“When you design recommendation systems,” she says, “you’re shaping what users believe is important.” Repetitive content risks creating echo chambers, subtly teaching users that certain topics dominate their reality.

Platforms are also designed to keep you engaged without any friction or interruption, so the experience feels seamless. But when you get annoyed, such as seeing irrelevant content, it creates friction. Platforms try to detect this ‘annoyance’ fast by watching for signals (like skipping content) so they can adjust their recommendations and keep you engaged.

But Dr. Piotrowksi challenges this idea of frictionless experiences: “Most systems relying on algorithms prioritize frictionless experiences in order to make sure the content flows easily. But a little friction can be beneficial. It gives users a moment to reflect, reconsider, or even disengage – and taking a break is not a bad thing.”

Three principles for healthier feed design

1. Limit reinforcement cycles

Rather than simply varying repeated content, designers should consider how to interrupt overly reinforcing loops. Even diverse content can contribute to cognitive fixation or emotional fatigue if it revolves around the same themes. Breaking these cycles — by deliberately introducing unrelated topics or even contrasting viewpoints — can be a more effective step toward building healthier, more balanced feeds.

Dr. Piotrowski notes that Dr. Amber van der Wal’s forthcoming work, Turning Algorithms for Good: ‘Injecting’ Humor to Protect Youth from Harmful Social Media Content, will be particularly relevant to this discussion.

2. Healthy defaults

Default settings matter. They set the baseline experience. Dr. Piotrowski notes, “We can’t assume users will actively curate their feeds. The majority simply accept defaults. So it’s our responsibility to ensure these defaults are diverse, balanced, and supportive of wellbeing.”

Defaults that promote variety and balance make it easier for all users, especially young people, to encounter more diverse content without too much extra effort. By designing defaults that are protective rather than purely engagement-driven, platforms can do more to prevent overexposure to narrow ideas.

3. Meaningful controls

Users need clear, accessible ways to manage their experiences. Dr. Piotrowski suggests providing options to pause, hide, or adjust recommendations. But she stresses these tools should empower rather than frustrate.

These controls shouldn’t be buried behind layers of settings, and they should be easy to understand and use in the moment.

Reclaiming agency over your feed

Taking back control of your social media experience doesn’t require deep technical knowledge, just intentional actions.

“Algorithmic literacy shouldn’t be optional,” Dr. Piotrowski argues. “Users, particularly young people, need tools and education to recognize when they’re being subtly guided or manipulated by their feeds. If you’re unaware, that’s more concerning.”

She suggests three ways that individuals can start making choices that reshape what they see every day.

1. Actively diversify your interests

Engage intentionally with varied content and creators, teaching the algorithm broader preferences. This might mean deliberately following accounts that share diverse perspectives or occasionally exploring content outside your usual interests. Over time, this variety can disrupt the narrow loops that automated systems tend to reinforce.

2. Use available curation tools

Regularly mute or hide repetitive or draining content. Most platforms now offer settings to remove specific types of posts or recommend fewer similar items. Dr. Piotrowski suggests treating these tools as essential hygiene rather than optional extras. Regularly pruning your feed helps prevent fatigue and leads to healthier engagement.

3. Talk about what you’re seeing

Talking openly, especially with young people, builds resilience and critical thinking, helping contextualize online messages and countering the perception that a repeated message reflects reality. Dr. Piotrowski notes that conversations about online content can build more resilience and develop critical thinking. By discussing what you see — and how often you see it — you create space to question and contextualize the messages being amplified.

The road ahead

As AI-generated content proliferates and algorithms grow smarter, the urgency of embracing algorithmic dispersion grows clearer. “We know that constant exposure to narrow content skews individual perspectives, but it also impacts social cohesion,” Dr. Piotrowski warns. “If everyone’s reality is intensely personalized, we lose common ground, making societal dialogue more challenging.”

But algorithmic dispersion offers a path forward: one that values breadth over clicks and balance over stickiness. For Trust & Safety teams, this means making the case that healthier feeds are better not just for compliance, but for the long-term health of communities.

For users, it means taking more control by diversifying interactions, managing settings, and maintaining awareness. But it also means recognizing that these are not intuitive skills. Learning to navigate algorithmic environments takes support, not just personal effort.

As a society, we have a responsibility to foster digital competence, through education, transparency, and shared norms, to ensure everyone has a fair chance at a healthier experience online.

WebPurify and IntouchCX help leading platforms combine human moderation and AI tools to build safer, more intentional online experiences. Learn more about our approach to Trust & Safety here.